Skip to main content

Table 5 Evaluating the proposed network in comparison with the state-of-the-art models tested by Jain et al. [48] for the detection of anomalies in the Kvasir-Capsule dataset

From: Wireless capsule endoscopy multiclass classification using three-dimensional deep convolutional neural network model

Model architecture

3D-CNN

2D-CNN

Simple 2D-CNN*

Weakly supervised CNN + iterative cluster unification*

VGG19 + InceptionV3 + ResNet50*

Dilated CNN*

Meta-feature parallel CNN*

Proposed by

This study

This study

Jia et al. [14]

Iakovidis et al. [49]

Caroppo et al. [12]

Goel et al. [50]

Hybrid CNN [48]

Sensitivity (%)

97

92

92

91

93

93

97

PPV (%)

98

89

87

89

96

93

97

F1 score

98

92

89

90

95

93

97

Accuracy (%)

98

93

90

91

95

93

97

  1. * The evaluation metrics are reproduced from the original study of Jain et al. [48]